Data representations and architectures, systems, and methods for multi-sensory fusion, computing, and cross-domain generalization

Abstract

A computer-implemented method, computer system and machine readable medium. The method includes performing a set of parameterizations of a plurality of semantic concepts, each parameterization of the set including: receiving existing data at a computer system on the plurality of semantic concepts, the existing data including processed output data from a plurality of neural network-based computing systems (NNBCSs), the processed output data corresponding to a plurality of distinct data domains associated with respective ones of the NNBCSs; generating a data structure to define a continuous vector space of a digital knowledge graph (DKG) based on the existing data; and storing the data structure in the memory circuitry; and in response to a determination that error rates from a processing of data sets by the plurality of NNBCSs are below respective predetermined thresholds, generating a training model.

1 - 25 . (canceled) 
     
     
         26 . A computer-implemented method of generating a training model regarding a plurality of semantic concepts, the method including:
 performing a set of parameterizations of the plurality of semantic concepts, each parameterization of the set including:
 receiving existing data at a computer system on the plurality of semantic concepts, the computer system including memory circuitry and a processing circuitry coupled to the memory circuitry, the existing data including processed output data from a plurality of neural network-based computing systems (NNBCSs), the processed output data corresponding to a plurality of distinct data domains associated with respective ones of the NNBCSs; 
 generating a data structure to define a continuous vector space of a digital knowledge graph (DKG) based on the existing data, the continuous vector space integrating the processed output data from the plurality of NNBCSs; and 
 storing the data structure in the memory circuitry; and 
   in response to a determination that error rates from a processing of data sets by the plurality of NNBCSs are above respective predetermined thresholds, performing a subsequent parameterization of the set, and otherwise generating the training model corresponding to the data structure from a last one of the set of parameterizations, the training model to be used by the NNBCSs to process further data sets.   
     
     
         27 . The computer-implemented method of  claim 26 , wherein performing a subsequent parameterization of the set includes generating the data structure by using a backward propagation of learning errors from each of the NNBCSs throughout a data structure of the DKG that led to the learning errors. 
     
     
         28 . The computer-implemented method of  claim 26 , wherein receiving includes receiving processed output data simultaneously from the plurality of NNBCSs. 
     
     
         29 . The computer-implemented method of  claim 26 , further including sending fused output data into a fused data NNBCS, the fused output data based on data fused from processed output data from the plurality of NNBCS. 
     
     
         30 . The computer-implemented method of  claim 26 , wherein the existing data further includes empirical data, the method further including receiving the empirical data at the computer system. 
     
     
         31 . The computer-implemented method of  claim 26 , wherein the plurality of NNBCSs are coupled to the memory circuitry, the method comprising using each of the plurality of NNBCSs to:
 access the training model in the memory circuitry; and   process, based on the training model, a respective data set from a respective one of the plurality of distinct data domains to generate a processed output data corresponding to the respective one of the plurality of distinct data domains.   
     
     
         32 . The computer-implemented method of  claim 31 , further including using at least one of processed output data corresponding to the respective one of the plurality of distinct data domains as part of the existing data set to perform a subsequent parameterization. 
     
     
         33 . The computer-implemented method of  claim 31 , the method including operating the neural network-based computing systems in parallel with one another to simultaneously process the respective data set from the respective one of the plurality of distinct data domains. 
     
     
         34 . The computer-implemented method of  claim 26 , further including, after storing the data structure and prior to performing the subsequent parameterization or generating the training model:
 receiving additional processed output data from an additional NNBCS, the processed output data corresponding to a plurality of distinct data domains associated with respective ones of the NNBCSs;   modifying the data structure based on the additional processed output data to generate a modified data structure defining a modified continuous vector space of the digital knowledge graph (DKG), the modified continuous vector space integrating the processed output data from the plurality of NNBCSs and the additional processed output data from the additional NNBCS; and   storing the modified data structure in the memory circuitry.   
     
     
         35 . The computer-implemented method of  claim 34 , wherein:
 the data structure corresponds to a Distributed Knowledge Graph (DKG) defined by a plurality of nodes each representing a respective one of the plurality of semantic concepts, the plurality of semantic concepts being based at least in part on the existing data, each of the nodes represented by a characteristic distributed pattern of activity levels for respective meta-semantic nodes (MSNs), the MSNs for said each of the nodes defining a standard basis vector to designate a semantic concept, wherein standard basis vectors for respective ones of the nodes together define the continuous vector space; and   each MSN corresponds to an intersection of a plurality of dimensions, each activity level in the pattern of activity levels designating a value for a dimension of the plurality of dimensions.   
     
     
         36 . The computer-implemented method of  claim 35 , wherein a dimension of the plurality of dimensions corresponds to a time dimension, and wherein an activity level for the time dimension represents one of time from a linear lunar calendar, time related to an event, time related to a linear scale, time related to a log scale, a non-uniform time scale, or cyclical time. 
     
     
         37 . The computer-implemented method of  claim 35 , wherein a dimension of the plurality of dimensions corresponds to a space dimension, and wherein an activity level for the space dimension represents one of linear scaled latitude, linear scaled longitude, linear scale altitude, building coordinate codes, allocentric polar coordinates, Global Positioning System (GPS) coordinates, or indoor location WiFi based coordinates. 
     
     
         38 . A computer system including a memory circuitry, and processing circuitry coupled to the memory circuitry, the processing circuitry including one or more input/output interfaces, the memory circuitry loaded with instructions, the instructions, when executed by the processing circuitry, to cause the processing circuitry to perform operations comprising:
 performing a set of parameterizations of the plurality of semantic concepts, each parameterization of the set including:
 receiving existing data at the one or more input/output interfaces of the computer system on a plurality of semantic concepts, the existing data including processed output data from a plurality of neural network-based computing systems (NNBCSs), the processed output data corresponding to a plurality of distinct data domains associated with respective ones of the NNBCSs; 
 generating a data structure to define a continuous vector space of a digital knowledge graph (DKG) based on the existing data, the continuous vector space integrating the processed output data from the plurality of NNBCSs; and 
 storing the data structure in the memory circuitry; and 
   in response to a determination that error rates from a processing of data sets by the plurality of NNBCSs are above respective predetermined thresholds, performing a subsequent parameterization of the set, and otherwise generating the training model corresponding to the data structure from a last one of the set of parameterizations, a training model to be used by the NNBCSs to process further data sets.   
     
     
         39 . The computer system of  claim 38 , wherein performing a subsequent parameterization of the set includes generating the data structure by using a backward propagation of learning errors from each of the NNBCSs throughout a data structure of the DKG that led to the learning errors. 
     
     
         40 . The computer system of  claim 38 , wherein receiving includes receiving processed output data simultaneously from the plurality of NNBCSs. 
     
     
         41 . The computer system of  claim 38 , the operations further including sending fused output data into a fused data NNBCS, the fused output data based on data fused from processed output data from the plurality of NNBCS. 
     
     
         42 . The computer system of  claim 38 , wherein the existing data further includes empirical data, the operations further including receiving the empirical data at the computer system. 
     
     
         43 . The computer system of  claim 38 , further including the plurality of NNBCSs coupled to the memory circuitry, the operations comprising using each of the plurality of NNBCSs to:
 access the training model in the memory circuitry; and   process, based on the training model, a respective data set from a respective one of the plurality of distinct data domains to generate a processed output data corresponding to the respective one of the plurality of distinct data domains.   
     
     
         44 . The computer system of  claim 43 , the operations further including using at least one of processed output data corresponding to the respective one of the plurality of distinct data domains as part of the existing data set to perform a subsequent parameterization. 
     
     
         45 . The computer system of  claim 38 , the operations including operating the neural network-based computing systems in parallel with one another to simultaneously process the respective data set from the respective one of the plurality of distinct data domains. 
     
     
         46 . The computer system of  claim 38 , the operations further including, after storing the data structure and prior to performing the subsequent parameterization or generating the training model:
 receiving additional processed output data from an additional NNBCS, the processed output data corresponding to a plurality of distinct data domains associated with respective ones of the NNBCSs;   modifying the data structure based on the additional processed output data to generate a modified data structure defining a modified continuous vector space of the digital knowledge graph (DKG), the modified continuous vector space integrating the processed output data from the plurality of NNBCSs and the additional processed output data from the additional NNBCS; and   storing the modified data structure in the memory circuitry.   
     
     
         47 . The computer system of  claim 38 , wherein:
 the data structure corresponds to a Distributed Knowledge Graph (DKG) defined by a plurality of nodes each representing a respective one of the plurality of semantic concepts, the plurality of semantic concepts being based at least in part on the existing data, each of the nodes represented by a characteristic distributed pattern of activity levels for respective meta-semantic nodes (MSNs), the MSNs for said each of the nodes defining a standard basis vector to designate a semantic concept, wherein standard basis vectors for respective ones of the nodes together define the continuous vector space; and   each MSN corresponds to an intersection of a plurality of dimensions, each activity level in the pattern of activity levels designating a value for a dimension of the plurality of dimensions.   
     
     
         48 . A product comprising one or more tangible computer-readable non-transitory storage media comprising computer-executable instructions operable to, when executed by at least one computer processor of a computer system including memory circuitry coupled to the at least one computer processor, enable the at least one processor to:
 perform a set of parameterizations of a plurality of semantic concepts, each parameterization of the set including:
 receiving existing data on the plurality of semantic concepts, the existing data including processed output data from a plurality of neural network-based computing systems (NNBCSs), the processed output data corresponding to a plurality of distinct data domains associated with respective ones of the NNBCSs; 
 generating a data structure to define a continuous vector space of a digital knowledge graph (DKG) based on the existing data, the continuous vector space integrating the processed output data from the plurality of NNBCSs; and 
 storing the data structure; 
   in response to a determination that error rates from a processing of data sets by the plurality of NNBCSs are above respective predetermined thresholds, perform a subsequent parameterization of the set; and   in response to a determination that error rates from a processing of data sets by the plurality of NNBCSs are below respective predetermined thresholds, generate a training model corresponding to the data structure from a last one of the set of parameterizations, the training model to be used by the NNBCSs to process further data sets.   
     
     
         49 . The product of  claim 48 , wherein performing a subsequent parameterization of the set includes generating the data structure by using a backward propagation of learning errors from each of the NNBCSs throughout a data structure of the DKG that led to the learning errors. 
     
     
         50 . The product of  claim 48 , wherein the plurality of NNBCSs are coupled to the memory circuitry, the instructions to enable the at least one processor to use each of the plurality of NNBCSs to:
 access the training model in the memory circuitry; and   process, based on the training model, a respective data set from a respective one of the plurality of distinct data domains to generate a processed output data corresponding to the respective one of the plurality of distinct data domains.